This week on Keep Going, I sat down with Scott Stevenson, CEO of Spellbook, a company that started with a frustrating legal bill and turned into one of the most widely used AI tools in the legal world. Scott’s story isn’t a straight line. He started building video games. Then musical instruments. None of them took off. But what stuck with him was the pain of trying to build something creative and getting hit with sky-high legal costs.

Instead of walking away from that experience, Scott leaned into it. He co-founded a company called Rally that evolved into Spellbook—now used by over 3,000 legal teams in more than 50 countries. If you've ever used GitHub Copilot, Spellbook is basically that for lawyers. It handles the repetitive stuff, helps draft documents, and is now expanding into what Scott calls "AI colleagues"—agents that can review and manipulate whole sets of legal documents across a transaction.

What stood out in our conversation was Scott’s willingness to chase hard problems. He said the turning point came when they stopped trying to make something flashy and started looking for work people actually needed. He and his team launched more than 100 product variations before finding one that clicked. They didn’t overspend. They didn’t force the market. They waited for the moment when users started grabbing the product out of their hands. And when it came, it came fast—30,000 signups in three months.

We also talked about art, creativity, and the tradeoffs that come with trying to make something lasting. Scott still wants to make music. Still wants to build games. But for now, he’s building tools that just might change how the legal system works—from bloated contracts to data-driven negotiations.

If you’re stuck, if you’re wondering whether the thing you’re building is really worth it, Scott’s advice is simple: Find a real problem. Stick with it. Wait for the water to start moving fast. You’ll know it when you see it.